Update app.py

app.py

@@ -3,12 +3,13 @@ import pandas as pd
 from sklearn.ensemble import RandomForestClassifier
 import joblib
 import gradio as gr
+import google.generativeai as gai
 
 # Load the trained classifier model
 model = joblib.load('model_pkl')
 
-# OpenAI API key (assumed to be set in Hugging Face environment)
-openai.api_key = 'sk-hVMYjAtIAwnqbMnArtTm9dgt7fC_snc4P7On7AtxebT3BlbkFJPVg50uJ3R6G4O6_2l6IUqLKtRC3bn_VQBIt1S9HtAA'
+
+gai.configure(api_key='AIzaSyAwP55Zlq9KqUBjHWWUjfzHcP4Sr8DVMuk')
 
 # Function to simulate the medical assistant's interaction
 def medical_assistant_interaction(pulse_rate, blood_pressure_systolic, blood_pressure_diastolic, temperature_celsius, 
@@ -49,16 +50,15 @@ def medical_assistant_interaction(pulse_rate, blood_pressure_systolic, blood_pre
 
         The patient reported the following symptoms: {patient_responses}.
         
-        Based on these readings and symptoms, what is the likelihood of Lassa fever? Provide additional follow-up questions if necessary.
+        Based on these symptoms, what is the likelihood of Lassa fever? Provide additional follow-up questions if necessary.
         """
 
-        # Call OpenAI's API for a response
-        response = openai.Completion.create(
-            engine="gpt-3.5-turbo",
-            prompt=response_text,
-            max_tokens=150
+        response = gai.chat(
+            model="chat-bison-001",  
+            messages=[{"role": "user", "content": response_text}],
+            max_output_tokens=150
         )
-        assistant_message = response['choices'][0]['text'].strip()
+        assistant_message = response.last['content'].strip()
 
         # Combine the assistant's message with the model's prediction
         if prediction == 1: